See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/337827374

Mathematical Modelling of Weld Phenomena 12 1 PREDICTION OF STRESS

CORROSION CRACKING IN 304 STAINLESS STEEL CANISTERS IN DRY
STORAGE OF SPENT FUEL BY MODELING: ANALYSIS OF WELD RE...

Article · December 2019

DOI: 10.3217/978-3-85125-615-4-12

CITATIONS READS

0 45

4 authors, including:

Scott Gordon Stephen Liu

Colorado School of Mines Colorado School of Mines
2 PUBLICATIONS 2 CITATIONS 67 PUBLICATIONS 437 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Conference paper View project

Reaction synthesis aluminum metal matrix composites produced via additive manufacturing View project

All content following this page was uploaded by Scott Gordon on 08 December 2019.

The user has requested enhancement of the downloaded file.

 Mathematical Modelling of Weld Phenomena 12

PREDICTION OF STRESS CORROSION

CRACKING IN 304 STAINLESS STEEL
CANISTERS IN DRY STORAGE OF SPENT
FUEL BY MODELING: ANALYSIS OF WELD
RESIDUAL STRESS AND SUSCEPTIBLE
MICROSTRUCTURE

S. GORDON*, X. WU*, Z. YU* and S. LIU*

*George S. Ansell Department of Metallurgical and Materials Engineering
Center for Welding, Joining and Coatings Research
Colorado School of Mines, Golden, CO 80401, U.S.A.

DOI 10.3217/978-3-85125-615-4-12

ABSTRACT
Atmospheric chloride-induced stress corrosion cracking (CISCC) in the weldments of spent fuel canisters
is one of the primary safety concerns during the dry storage of used nuclear fuel at Independent Spent Fuel
Storage Installations (ISFSI) in coastal areas. For SCC to occur, three criteria must be met: an aggressive
chemical environment, susceptible microstructure, and sufficient tensile stress. Instead of the
environment, this paper will focus only on the material microstructure and stress state. Firstly, finite
element analysis (FEA) based numerical simulation models were developed for the welding of the mockup
canister. Multi-pass arc welding process was considered. The thermal history and residual stress
distribution induced by the longitudinal, circumferential, and the intersection between longitudinal and
circumferential welds were developed based on a thermal-mechanical coupling model using ABAQUS
software. The simulation results were compared with the deep-hole drilling (DHD) and contour
measurement data. Based on the model-predicted stress results, a four-point bend (4PB) setup was then
designed to duplicate the weld residual stress onto the 4PB specimen, on which the maximum tensile stress
is close to the predicted maximum stress level of 250 MPa in the weld heat-affected zone (HAZ). A stress
gradient also existed along the thickness of the 4PB specimen which allowed for the consideration of
stress variation. The designing of the 4PB test was done by FEA method using ABAQUS software, and
verified by digital image correlation (DIC) measurements. Besides, to identify the most susceptible
microstructure of the canister to corrosion under the weld residual stress in the controlled environment,
the modified implant test (CEMIT) was designed. A FEA model considering the welding process and the
following uniaxial tensile load was then developed using the SYSWELD software. The residual stress and
thermal history were obtained during and after the CEMIT welding process, and after the uniaxial tensile
loading process. These tests yielded instructive information for the prediction of the initiation site of the
crack. The initiation of the crack observed in the experiments was explained based on the simulation data.
While each of the three simulations provided an independent story, the three together produced a complete

1
 Mathematical Modelling of Weld Phenomena 12

and complementary picture. Stress corrosion cracking is a series of complex interactions which
simultaneously occur. The thermal mechanical model of the canister provides information for predicting
the location at which stress corrosion cracking will most likely occur. The model for the 4PB test predicts
the stress gradient which occurs in a bent bar. The SYSWELD CEMIT model predicts the welding cycle,
the thermal history, residual stress and loading behavior of the specimen. By design, each independent
CEMIT provides the location of the most crack susceptible location under a specified environment.

Keywords: Stress corrosion cracking, Spent fuel canister, Weld residual stress, Susceptible microstructure,
Modelling

INTRODUCTION

CISCC in the weldments of spent nuclear fuel (SNF) canisters is one of the primary safety
concerns during the dry storage of used nuclear fuel at Independent Spent Fuel Storage
Installations (ISFSI) in coastal areas [1]. For SCC to occur, three criteria must be met: an
aggressive chemical environment, susceptible microstructure, and sufficient tensile stress.
Field sampling analysis of surface deposits on in-service spent nuclear fuel storage canisters
at three near-marine ISFSI sites in the United States have demonstrated the presence of
chloride-rich salts on the canister surfaces [2-4]. As part of the surface of the canisters cools
sufficiently for the salts to deliquesce, chloride-rich aqueous brine layer could form locally
on the surface, leading to potential failures by CISCC in regions with high tensile stress.
On the other hand, for austenitic stainless steel SNF interim canisters, the multi-pass
welding procedures introduce high tensile residual stresses and sensitization in the HAZ.
Therefore, the potential for CISCC in the weldments of the stainless steel interim storage
canisters has been identified as a high priority data gap.
To gain fundamental understanding of the pitting and cracking behavior in the
canisters and to accurately predict their long-term performance, the first step is to obtain
the accurate residual stress distribution within the canisters, especially in the weldments. In
the last decade, there have been several studies on the residual stresses in the canister welds.
For example, Kosaki et al. [5] reported that the experimentally measured weld residual
stresses were close to the yield stress through thickness on a cylindrical canister of 1.3 m
in diameter and with walls approximately 75 mm thick. The Nuclear Regulatory
Commission (NRC) simulated the residual stresses for typical canister welds by employing
a two-dimensional, sequentially-coupled thermal-structural FE model [6], and predicted
that tensile stresses of sufficient magnitude to initiate SCC are likely to be present in the
HAZ of both longitudinal and circumferential welds through the wall thickness of the
canisters, allowing for crack propagation through the wall thickness over time. There is
very limited literature on 3D simulation of the residual stress induced by multi-pass
longitudinal and circumferential welds in the dry storage fuel canisters. As for the
microstructure of the canister material, possible factors that may influence the CISCC
include the sensitization, ferrite, martensite, inclusions, grain size, phase orientation,
texture, and so on. It is, thus, necessary to determine the most susceptible microstructure
for the weld region of the canister under the influence of weld heat input and mechanical
stresses. In this study, a 3D FE model using ABAQUS software was built to simulate the
residual stress contours in the 304 stainless steel canisters with the presence of both
longitudinal and circumferential multi-pass welds. Experimental measurements by deep
hole drilling (DHD) and contour methods provide valuable comparisons with the simulation

2
 Mathematical Modelling of Weld Phenomena 12

results. Then, a modified four-point bend setup was built to duplicate the weld residual
stresses into a small sample, by which subsequent corrosion tests will be carried out. At
last, a modified implant test was designed, together with FEA, to characterize the most
susceptible microstructure for the initiation of SCC.

SIMULATION DETAILS AND EXPERIMENTAL METHODOLOGIES

WELD RESIDUAL STRESS SIMULATION

The ABAQUS software was used to build the residual stress models for longitudinal and
circumferential double-V groove multi-pass welds. Fig. 1(a) demonstrates the meshed
structure with 52100 brick elements established based on the actual geometry of the
mockup and the welding conditions as schematically illustrated in Fig. 1(b). The mockup
consists of three cylindrical shells that were 4 feet (~1220 mm) long each and 5/8 inch
(~15.9 mm) thick as shown in Fig. 1(b). The shells were rolled from a plate into a cylinder
and then welded at the longitudinal seam. The three cylinders were joined together by two
circumferential welds. Details of the welding parameters including pass sequence, voltage,
current, travel speed and heat input for each weld pass can be found in reference [7]. Take
longitudinal welds for example, the three inner passes were made first, followed by the four
passes in outer diameters. Since the three longitudinal welds were arranged 180 degrees
apart (see Fig. 1(b)) and the two circumferential welds were separated by the middle
section, the interference between the same welding types can be negligible. Hence, in this
FE model, only the longitudinal and circumferential intersecting joint in half of the mockup
section was simulated, as shown in Fig. 1(a), and the welding length was sufficient to reach
steady state. Note that the heat source travelled through the whole length of the welds to
ensure the elements experience the same thermal history as did the actual welds.

3
 Mathematical Modelling of Weld Phenomena 12

Fig. 1 (a) The FEA model for the mockup with fine meshes near welds, and (b) schematic
drawing of the mockup.
The multi-pass weld beads have sharp transition angles, as shown in Fig. 2, which
demands extremely fine meshes. More importantly, the intersecting longitudinal and
circumferential welds needs compatible meshing. Due to such complications, the weld bead
was divided into small rectangle-shaped areas matching the experimental bead morphology
as closely as possible. The experimental bead morphology of each pass is outlined by the
dotted curves in Fig. 2 and the corresponding simulation passes are highlighted in various
colors. Such simplification avoided elements with sharp angles and assisted the FE model
to converge more easily.

Fig. 2 Experimental bead morphology and the corresponding computational weld bead. In the
figure, “in” means inner diameters of the mockup, and “out” refers to the outer diameters of
the mockup.
The mockup base material is 304 stainless steel (SS). The material properties used in the
simulation were taken from Deng’s work [8]. Temperature-dependent hardening modulus
used in this model was obtained from Liu’s work [9]. The material solidus and liquidus
temperatures were set at 1400 and 1450 ℃, respectively. The latent heat of fusion was set
to be 260 kJ/kg. The material properties were assumed to be the same for both base and
weld metals due to lack of information in 308 weld metal physical properties.
Double-ellipsoidal heat source model developed by Goldak [10] as illustrated in Fig. 3
was used to simulate the heat input of submerged-arc welding process. The front and rear
parts of the heat flux are descried by Eqs. (1) and (2), respectively:

6 3 f r Q 3x 2 3 y 2 3z 2
q（x, y, z）= exp(− − 2 − 2 )
ar bc 
r
ar 2 b c
(1)

6 3 f f Q 3x 2 3 y 2 3z 2
q（x, y, z）= exp(− 2 − 2 − 2 )
a f bc 
f
af b c
(2)

where the front and rear quadrant fractions, ff and fr, were set to be 0.6 and 1.4,
respectively. The heat input was defined as Q = V*I*η, where V = 30 V, I = 400 A and η =
0.8. The constants af, ar, b, and c were set to be 13, 26, 13, and 13, respectively, to closely
match the experimental fusion boundary in Fig. 2.

4
 Mathematical Modelling of Weld Phenomena 12

Fig. 3 Heat source model for the weld simulation.

The weld residual stresses were also characterized by DHD and contour measurement
method developed by Sandia National Laboratory [7], which can be used to validate the
simulation data. Detail comparison between simulation and experimental data are discussed
in Section 3.

REPLICATION OF RESIDUAL STRESSES BY FOUR-POINT BENDING TEST

The stresses experienced by the mockup materials were duplicated to laboratory scale
specimens through four-point bend (4PB) test. In this modified 4PB test setup, pitting
initiation/growth was characterized within the thickness of the specimen, which covered
stress levels from compression to tension. The maximum tensile residual stress, as indicated
in another report by the same authors [11], was predicted to be around 250 MPa in the HAZ,
so the design of this 4PB setup was to obtain maximum stress on the tensile side close to
250 MPa. FE method was adopted to design the geometry of this setup to allow for a
sufficient sampling area within a small stress increment. Fig. 4 illustrates the finalized
geometry design. The frame and actuator material is 316 stainless steel, and they were
coated with Kolor-Proxy high build polyamide-epoxy primer that provides abrasion, impact
and chemical resistance.

5
 Mathematical Modelling of Weld Phenomena 12

Fig. 4 Optimized geometry of the modified 4PB setup by FEA.

The stress loading of the specimen was accomplished by turning the two screws
downward to certain displacement predicted by FEA. DIC method was used to in-situ
monitor the strain development in the designed 4PB setup during incremental loading for
the purpose of validation. Fig. 5(a) shows the DIC measurement setup. Fig. 5(b)
demonstrates the surface preparation for DIC measurement, where a white background and
black speckle pattern was generated by carefully spraying the white coating primer and low
chloride metal marker black paint.

Fig. 5 (a) Photograph of the DIC setup for 4PB testing, and (b) the speckle pattern on the side
surface of 4PB setup prepared for DIC analysis.

MODIFIED IMPLANT TEST

The following discussion will show how the Controlled Environment Modified Implant
Test (CEMIT) can correlate the weld thermal exposure with the developed microstructure
in the heat affected zone (HAZ) of the welds as well as the sensitization experienced.
Together with the stresses developed during welding and subsequent canister handling, the
sensitized microstructure consisted of austenite with precipitated Cr23C6 particles and small

6
 Mathematical Modelling of Weld Phenomena 12

amounts of deformation-induced martensite contributes to stress corrosion in the corrosive

environment.
As a result of the high temperature, the base material near the fusion line may recover
and/or recrystallize, and if near enough to the weld zone, may melt. Softening generally
occurs in this region. Further away from the weld there is a region where the kinetics for
carbide (Cr23C6) precipitation is favorable. The material within this region is heat treated to
between 450 to 850oC and is commonly known as the sensitized region and under proper
conditions can be more prone to corrosion. The upper and lower limits of the temperature
range can change according to the chemical composition, in particular, carbon content, of
the stainless steel. Alloys with lower carbon content will see lower temperatures and longer
delay times for carbide precipitation. To account for residual stress development, a thermal
mechanical model was developed to simulate the conditions in the canister during and after
welding. Due to the size of the canister, the complex residual stress states, and ever
changing natural environment, determining the exact location and time when SCC will
happen is difficult.

Fig. 6 A threaded cylindrical specimen with a weld bead on top. A tensile load is applied at the
conclusion of the welding. The arrow indicates the direction of loading.
The Implant testing was substantially modified from its original intention [12-13] with
the addition of an environmental chamber to introduce the corrosive elements (salt load,
relative humidity, and temperature). To simulate the environment of the canister, salts
would be deposited in the threaded region of the welded cylinder and the specimen would
be placed into the chamber with preset temperature and humidity. The specimen would then
be loaded to a pre-determined stress level. Under load, the notch concentrates the stress to
the root of the threads. Fig. 6 shows a generalized schematic drawing depicting the basic
concept of the implant test. When the specimen is provided with the proper conditions for
corrosion, pitting and cracking can be observed in the notched region.
A thermal mechanical model using SYSWELD was developed to simulate the welding
process for the Implant test. The thermal histories determined from both models: the
canister and the Implant test can be compared. Similar thermal cycles found in the models
should produce similar microstructures in both weldments (on the canister and in the
Implant test). By using the thermal cycle determined from the Implant test model at the

7
 Mathematical Modelling of Weld Phenomena 12

location in the threads where pitting corrosion and cracking occurred, a corresponding
location for crack-susceptible microstructure can be determined on the model developed
for the canister. Grain size as well as degree of sensitization can be used to support the
microstructural behavior prediction.
By using the temperature-time profiles determined by the models, bulk material with the
crack-susceptible microstructures can then be produced for testing by using the Gleeble or
other heat treatment techniques. Testing of this material can lead to the determination of
pitting initiation times, crack growth rates, and resistance to cracking once cracks are
initiated.
For atmospheric SCC testing, the basic Implant testing equipment would be modified.
However, the basic fundamentals of the test would remain the same, i.e. a welded threaded
specimen is pulled in tension. The major differences are, (1) after welding, the threaded
region of the specimen is coated in a corrosive media (salt loading), and (2) instead of
ambient conditions, the tensile loading of the specimen is conducted under a controlled
environment. By loading the end of this vertical cylindrical specimen of Type 304 stainless
steel in the corrosive media, the spiral notch will locate the position of the microstructure
most susceptible to cracking.

Implant specimen

The Implant specimen is a Type 304 stainless steel cylinder machined with a spiral notch
as shown in Fig. 7(a). Fig. 7( b) is a schematic drawing showing the orientation of the weld
cup and implant specimen with respect to rolling direction of the base plate. The specimen
would be machined from 0.625 inch thick plate transverse to the direction of rolling. The
specimen is approximately 3.5 inches long with a ¼-28 thread on one end and ½-13 thread
on the other. The end with the ¼-28 thread was the welded end and the other end connected
to the load cell. Table 1 shows the chemical composition of the implant specimen.

(a) (b)
Fig. 7 (a) Photograph showing Implant specimens and adapter collar screwed onto the bottom
of the specimens. (b) Schematic drawing showing the implant specimen and weld cup
extraction location with respect to base plate orientation.
Table 1 Chemical composition (in weight percent) of Type 304L stainless steel used in this
study

8
 Mathematical Modelling of Weld Phenomena 12

Thermal mechanical model

A commercial finite elements software, SYSWELD, was used to model the weld thermal
experience and stresses developed during the welding process. This software is also able to
predict the resulting weld microstructure as a function of the location on the modelled
component. Combining the thermal conditions data with the microstructural evolution
history in the specimen, bulk specimens can be created with the targeted microstructural
for corrosion susceptibility, pitting-to-cracking transition, fracture mechanics testing, and
advanced nondestructive evaluation. These specimens can also be further processed to
characterize in-situ corrosion and crack propagation mechanism.
Using as input the welding heat input, physical dimension and geometry of the material,
physical material properties, and applying the necessary thermal boundary conditions,
SYSWELD was used to model the temperature profiles, microstructure, mechanical
properties and the residual stresses developed in the cylindrical specimen.

Model description

Fig. 8 shows the meshed model used in the FEA. The Implant test specimen is constituted
of three parts: Base metal, weld metal and the screw. All the three parts are made of type
304 stainless steel; the material properties are chosen from reference [8]. The cross-section
view and top view as shown in Fig. 8(a) and Fig. 8(c) depict that the base metal (described
earlier also as weld cup) is modeled as a cylinder with a hole created in the middle. This
hole is simulated by a cylinder on the top and an inverted cone at the bottom. The screw is
connected with the base metal at the bottom of the cone. Welding is conducted on the top
of the screw and the weld metal is represented by the brown-colored region. As shown in
Fig. 8(d), the top 10 threads in the screw are meshed with very small size elements to clearly
describe the most sensitive region under the thermal and mechanical load. Different parts
are connected together by the co-node method. The thread is meshed with 4-node
tetrahedral elements due to the complicated geometry structure, while the rest is meshed
with hexahedral solid elements. The total number of elements is 754,374. The FEA studies
for the modified Implant test are conducted using the SYSWELD software.

Fig. 8 Meshed model for the modified Implant test simulation. (a) Cross-section view. (b) Front
view. (c) Top view. (d) Enlarged view.

9
 Mathematical Modelling of Weld Phenomena 12

Table 2 SYSWELD simulation parameters

Parameter Welding Welding Welding Welding Cooling Tensile
Current voltage velocity time (s) time (s) load
(A) (V) (m/s) (MPa)
Value 250 32 0.225 4.534 600 170

Fig. 9 Schematic drawing showing the progression of welding, cooling, and loading boundary
conditions.
The X, Y, and Z directions for the implant specimen shall be defined as the following.
The Y direction is the direction in which the specimen is a uniaxial loaded condition, the X
and Z directions are transverse to the loading direction. Fig. 9 describes the simulation
procedures used in SYSWELD. Firstly, spot arc welding is conducted on the top of the
screw; the spot welding is realized following a small circular trajectory. During the welding
process, the bottom surface of the screw is fixed. After welding, the three pieces are air
cooled for sufficient time. At this stage, the ambient temperature is kept as 20 0C with the
bottom surface of the screw fixed. Meanwhile, movement of the base metal in the -Y
direction is restricted. Then, after cooling, the whole piece is unclamped with three nodes,
during which no thermal or mechanical load is applied. Lastly, a uniaxial tensile load is
applied to the bottom surface of the screw. The tensile load value is adjusted to achieve a
stress level on the screw to be a fraction of the yield strength of the material (250 MPa).
During the loading process, movement of the base metal along the –Y direction is also
restricted. By taking the four simulation procedures, a process similar to that observed in
the experiment is reproduced on the computer. Table 2 gives the simulation parameters
used in SYSWELD. The choice of these parameters is also closely related to the real
conditions in experiment.
During the welding process three clamping conditions can arise. The first, both pin and
plate (weld cup) are free to expand and contract without restraint during welding or cooling.
Second, pin and plate are fixed during welding or cooling. Third, the pin and plate can both
move during welding but are fixed during cooling. The third case fits best the intended
testing condition. While the specimen is held by the fixture, the pin is free to expand through
the hole in the plate upon heating, and the plate is held by gravity to the fixture. Once the

10
 Mathematical Modelling of Weld Phenomena 12

melt pool incorporates both the pin and the plate, and solidification begins, the specimen is
deemed as fixed together. As the specimen begins to cool the weld metal has sufficient
strength to support the stress in the specimen.

RESULTS AND DISCUSSIONS

WELD RESIDUAL STRESS

Residual stress contours

Fig. 10 summarizes the overall stress contours in the mockup after seven longitudinal weld
passes and eight circumferential weld passes. The thermal contraction during solidification
and cooling of arc welding induced high tensile residual stress in the weld metals. In the
longitudinal welds, as shown in Fig. 10(a), the axial stress is as high as 423.7 MPa in the
weld center, which is much higher than the hoop stress. In comparison, a maximum residual
stress of 366.1 MPa lies in in hoop direction in the circumferential welds. Both stresses in
the two welds are way higher than yield strength (~ 250 MPa), so plastic deformation has
initiated. The influence of two intersecting multi-pass weldments on the residual stress
distribution is clearly demonstrated in Fig. 10. It should be noted that the highest tensile
stress always coincided with the welding direction.

Inner surface Outer surface

(a)

Inner surface Outer surface

(b)

Fig. 10 Overall residual stress contours for (a) axial stress and (b) hoop stress from FEA.

11
 Mathematical Modelling of Weld Phenomena 12

Comparison of FEA with measurement results

Fig. 11 summarizes the residual stress profiles as a function of depth in circumferential

weld centerline and HAZ (approximately 4mm away from weld toe on the outer surface)
obtained from FEA and DHD measurements, which demonstrates a good agreement. Note
that the experimental residual stress data close to the surfaces (≤2mm) were obtained from
conventional hole drilling method. Through-wall high hoop tensile stress was observed in
the circumferential welds. The maximum hoop stresses in circumferential welds are about
230 to 340 MPa in the weld centerline and approximately 200 MPa in HAZ, respectively.
The pre-existing residual stress in the base metal (up to 50MPa indicated by the DHD
measurements) induced by the rolling process during manufacturing which was not
considered in the FE model, may lead to the difference in residual stress values near
surfaces between experimental and simulation results. In comparison, for the longitudinal
welds (which is not shown here), high through-thickness axial stresses were observed in
the weld metal (320 to 420 MPa) and HAZ (170 to 200 MPa).
400 300
(a) (b)
250
300

200
Residual Stress (MPa)

Residual Stress (MPa)

Axial_DHD
200 Hoop_DHD
Axial_FEA 150 Axial_DHD
Hoop_FEA Hoop_DHD
100 Axial_FEA
100
Hoop_FEA

0 50

0
-100
-50
-2 0 2 4 6 8 10 12 14 16 18 -2 0 2 4 6 8 10 12 14 16 18
Depth from Outer Diameter (mm) Depth from Outer Diameter (mm)
Fig. 11 Residual stress profiles as a function of depth in (a) weld centerline and (b) HAZ of
circumferential weld from FE simulation (dashed lines) and DHD measurements (solid lines).
The data of FE simulation is averaged from ten points in the steady-state location.
The cross-section stress contour is extracted from FEA results, and compared with the
results from contour method for both circumferential and longitudinal weld. Fig. 12
demonstrates the comparison of longitudinal weld. Good agreement is achieved between
the FEA and contour measurement results. The residual stresses are strongly tensile through
thickness near the longitudinal weld centerline. The simulation result predicts a slightly
wider tensile residual stress regime than the contour method, especially on the inner and
outer surfaces, which is in agreement with the experimentally observed fusion boundary
morphology as illustrated in Fig. 2. Note that tensile stress is observed at the two ends of
the welded plate only in the contour method, as shown in in Fig. 12(a), which could be
introduced potentially by cutting process.

12
 Mathematical Modelling of Weld Phenomena 12

Fig. 12 Cross-section axial stress contour distribution for longitudinal weld from (a) contour
method and (b) FEA.

STRESS AND STRAIN DISTRIBUTION OF FOUR-POINT BEND SAMPLE

Verification of FEA results by DIC

The DIC method was able to provide the strain evolution along x axis (//LD, longitudinal
weld direction) as a function of displacement load along y axis (//ND, normal direction) in
a modified 4PB specimen The displacement load was carefully applied by turning the two
screws in Fig. 5(b) downward slowly and evenly. Good agreement was achieved between
the strain profiles obtained from DIC measurement and FEA, as plotted in Fig. 13. As
demonstrated by the stress profile, for every 2 mm increase along the thickness direction (y
axis), the variation of the stress load would be with a range ±25 MPa, meanwhile, the
constant stress region could extend to around 120 mm along x axis, which could result in a
sampling area as 2 mm×120 mm.

Fig. 13 Comparison of strain contours obtained from (a) FEA and (b) DIC measurement on the
modified 4BP specimen, and (c) the strain variations as a function of depth from top surface in
the center of the specimen subtracted from the strain maps in (a) and (b).

13
 Mathematical Modelling of Weld Phenomena 12

Stress and strain distributions for the 4pb samples

Fig. 14 gives the strain and stress distributions of the 4PB samples. The DIC method was
applied to track the strain evolution during the loading process so as to validate the FEA
design. As shown in Fig. 14(a), good agreement was achieved between the strain contour
obtained from FEA and DIC measurement. The stress contour was given in Fig. 14(b),
which indicates a stress variation from compression to tension along the thickness of the
sample. The maximum stress locates at the bottom surface of the sample, which is close to
250 MPa. The stress profile along the thickness in center (Fig. 14(c)) also demonstrates the
stress gradient.

Fig. 14 Strain and stress distribution of the loaded 4PB samples. (a) Comparison of the strain
field predicted by FEA and DIC. (b) The stress field predicted by FEA. (c) Stress profile along
path L. Only half of the specimen was shown due to symmetry.

THERMAL HISTORY AND SUSCEPTIBLE MICROSTRUCTURES

Fig. 15 is a temperature contour map generated by SYSWELD showing the specimen in

3D and a cross-section of the specimen at the end of the welding time. Fig. 15(b) is a graph
showing the temperature profile at the surface of each thread as a function of time. Fig.
16(a) is a photograph showing the root and cap bead of a completed CEMIT specimen
before testing. Fig. 16(b) shows a cross section of a tested specimen. The numbering of
the threads begins at the toe of the weld and increases as the distance increases, away from
the weld. The fusion line is annotated as (FL), weld metal as (WM), heat affected zone as
(HAZ), and base material as (BM), respectively. Where the base metal and the weld metal
meet on the surface, this location will be referred to as the toe of the weld. Since the weld
is actually a melt-through, this location will be referred to as the root of the weld. The fusion
line begins internally in the weld cup and exits towards the threads. Note the smooth wetting
angle between the threaded portion of the specimen and the weld cup. Fig. 17 is a series of
contour maps showing the development in strain and thermal experience. Note that tensile
stresses develop between the fusion line and Thread 8. Compression is observed after
Thread 8. By matching the actual

14
 Mathematical Modelling of Weld Phenomena 12

(a)

(b)

Fig. 15 (a) Temperature contour maps generated by SYSWELD. Graph showing temperature
profile at the surface of each thread as a function of time. (b) Simulated time-temperature
profiles for each thread generated by SYSWELD.
fracture location observed in the modified Implant test specimens with the SYSWELD
calculated temperature contours, the region where the microstructure was most sensitive to
cracking could be identified. Fig. 17(a) show the stress contour maps of the specimen upon
release from the fixture after cooling, with clear indication that the specimen is loaded
during testing. Note the regions in Threads 5 to 8 that does not stress to the same levels as
the upper portions of the threads. High surface and subsurface tension in the specimen close
to the fusion line may promote crack opening in this region if the microstructure is
predisposed to cracking. Thus, it is important to match the microstructure with the
developed stresses to confirm the effect of sensitization. Fig. 17(b) shows the stress contour
within the specimen after unloading and then the stress reapplied to 170 MPa. Note that
while loading the stress applied is below the yield stress, the stress levels within the first 7
threads reach a level at or above the yield strength due to the residual stresses. This is
important since the actual welds within the canister also have residual stress.

15
 Mathematical Modelling of Weld Phenomena 12

Top View Bottom View

Cap Bead Fusion Line/ Toe

a) b)

c)
Fig. 16 (a) Photographs showing a completed weld with acceptable appearance viewed from
both the top into the weld cup and from the bottom with threaded test region and weld root in
view. (b) Photograph showing the cross-section of a completed weld. The weld threads have
been numbered beginning at the fusion line and increasing in number as the base material is
approached. The weld, fusion line (FL), heat affected zone (HAZ) and base metal (BM) are
labelled. Where the base metal and the weld metal meet on the surface, this will be referred to
as the toe of the weld. The HAZ is between the fusion line and the base material. This is the
region where sensitization occurs. The BM is the material which is unaffected by any
processing.
Several implant specimens were produced and tested. Minor corrosion and cracking were
observed throughout all of the specimens, and in particular at the base of the threads, the
fusion line, and in regions where residual stress and cold work were produced during the
machining process. Observation showed that corrosion and SCC occurred in the HAZ,
between Threads 5 and 10, which correspond well with the region that experienced between
450 to 850 oC as predicted by the software. Fig. 18 shows two SEM images of the CEMIT
specimen. Fig. 18(a) is an image taken at 23x showing indications of corrosion in Threads
6 to 8. Fig. 18(b) is an SEM photograph taken from the cross-section of the same implant
specimen in Thread 7. A crack can be observed splitting the grain boundary between the
austenite (light color) and the delta Ferrite (darker color). Since the initial processing
conditions for wrought 304L includes a water quench, it is reasonable that small amounts
of delta ferrite remain in the base material. Using a Feritscope and SEM, it was confirmed
that ferrite was present. Small precipitates and carbides were observed decorating the grain

16
 Mathematical Modelling of Weld Phenomena 12

boundary. As shown in Fig. 15(b), Thread 7 reached a peak temperature just below 600 oC.
This condition was consistent with that needed for the nucleation and growth of the M23C6,
thus leading to sensitization and potentially, SCC. The formation of the M23C6 was verified
by performing a selective digestion mothed and then analysed using XRD. Further
verification of was performed by creating carbon extraction replicas and observing the
precipitates using TEM. The carbides were observed in the etch pits and decorating the
grain boundaries. It was found that the carbide morphology was consistent with literature
and EDS results were reasonable. While a kinetic model would help to show that
favourable conditions existed for the formation of the carbides, this was not the focus of
this study. Future work should and will include this type of model.

(a)
(b)
Fig. 17 (a) Contour maps showing the development in strain and thermal experience
(temperature history) along the screw. (b) Stress contour map showing the specimen after
welding (L) and upon release from the fixture (R) showing the stress developed upon reloading
to 170 MPa.

17
 Mathematical Modelling of Weld Phenomena 12

γ δ-

Carbides
and PPTs γ

Crack

(a) (b)
Fig. 18 (a) SEM image taken at 23X magnification showing corrosion on Threads 5 to 9 from
top to bottom. (b) SEM micrograph taken at 85000x showing a crack propagating through the
boundary between ferrite and austenite. Note the small carbides decorating the grain boundary.

CONCLUSIONS

From the numerical studies performed on the canister welding, 4PB test and CEMIT, it was
concluded that:
(1) Based on the thermal and stress results of the CEMIT predicted by the FEA model,
the crack initiation site observed in the microscope can be clearly explained. These
numerical studies could offer important information about the susceptible
microstructure and tensile stresses for the canister, which is meaningful for the
future investigation of the initiation and growth rate of CISCC.
(2) The canister simulation was validated by measured stress data obtained using deep-
hole drilling and contour methods. High through-thickness hoop stresses were
found within the weld metal (230 to 340 MPa) and HAZ (approximately 200 MPa)
in circumferential welds. In comparison, high through-thickness axial stresses
were observed in the weld metal (320 to 420 MPa) and HAZ (170 to 200 MPa) in
longitudinal welds.
(3) A 4PB test setup was reasonably designed by FEA and DIC to duplicate the stress
conditions in the mockup canister. For every 2mm increase along the thickness
direction, the variation of the stress load would be with a range ±25 MPa, which
could result in a sampling area as 2 mm×120 mm.
(4) Corrosion and cracking was observed in the CEMIT process concentrated within
threads 5 to 9. This region falls within the temperature range suitable for the
precipitation of the M23C6 carbides. The combination of this sensitized
microstructure, a sufficient load, and the corrosive environment lead to fracture in
the CEMIT specimen.

18
 Mathematical Modelling of Weld Phenomena 12

ACKNOWLEDGEMENTS

This work was supported by the DOE, Nuclear Engineering University Program (NEUP)
under grand number DE-NE0008442 of Integrated Research Program (IRP), no. IRP-15-
9318. The authors would like to gratefully acknowledge ESI Group for making the
SYSWELD software available to the researchers of the Center for Welding, Joining and
Coatings Research of the Colorado School of Mines. Their technical support and discussion
are also kindly appreciated.

REFERENCES

[1] R. M. KAIN: ‘Marine Atmospheric Stress Corrosion Cracking of Austenitic Stainless Steels’,
Materials Performance, Vol. 29, No. 12, pp. 60-62, 1990.
[2] D. G. ENOS, C. R. BRYAN and K. M. NORMAN: ‘Data Report on Corrosion Testing of Stainless
Steel SNF Storage Canisters’, FCRD-UFD-2013-000324, U.S. Department of Energy, Office
of Used Nuclear Fuel Disposition, 2013.
[3] C. R. BRYAN and D. G. ENOS: ‘Analysis of Dust Samples Collected from Spent Nuclear Fuel
Interim Storage Containers at Hope Creek, Delaware, and Diablo Canyon, California’,
SAND2014-16383, Albuquerque, NM, Sandia National Laboratories, 2014.
[4] C. R. BRYAN and D. G. ENOS: ‘Analysis of Dust Samples Collected from an Unused Spent
Nuclear Fuel Interim Storage Container at Hope Creek, Delaware’, SAND2015-1746,
Albuquerque, NM, Sandia National Laboratories, 2015.
[5] A. KOSAKI: ‘Evaluation method of corrosion lifetime of conventional stainless steel canister
under oceanic air environment’, Nuclear Engineering and Design, Vol. 238, No. 5, pp. 1233-
1240, 2008.
[6] NUCLEAR REGULATORY COMMISSION (NRC): ‘Finite Element Analysis of Weld Residual
Stresses in Austenitic Stainless Steel Dry Cask Storage System Canisters’, NRC Technical
Letter Report (ADAMS ML13330A512), Washington, D.C., 2013.
[7] D. G. ENOS and C. R. BRYAN: ‘Characterization of Canister Mockup Weld Residual Stresses’,
FCRD-UFD-2016-000064, SAND2016-12375 R, Albuquerque, NM, Sandia National
Laboratories, 2016.
[8] D. DEAN and M. HIDEKAZU: ‘Numerical simulation of temperature field and residual stress in
multi-pass welds in stainless steel pipe and comparison with experimental measurements’,
Computational Materials Science, Vol. 37, No. 3, pp. 269-277, 2006.
[9] C. LIU, J. X. ZHANG and C. B. XUE: ‘Numerical invetigation on residual stress distribution and
evolution during multi-pass narrow gap welding of thick-walled stainless steel pipes’, Fusion
Engineering and Design, Vol. 86, No. 4-5, pp. 288-295, 2011.
[10] J. GOLDAK, A. CHAKRAVARTI and M. BIBBY: ‘A new finite element model for welding heat
sources’, Metall Trans B, Vol. 15, No. 2, pp. 299-305, 1984.
[11] X. WU, Z. YU: ‘Final Report: Weld Residual Stress Analysis in the Canister Mockup’, DE-
NE0008442 of Integrated Research Program No. IRP-15-9318, Golden, CO, Colorado School
of Mines, 2018.
[12] H. GRANJON: ‘The Implants Methods for Studying the Weldability of High Strength Steels’,
Metal Construction and British Welding Journal, Vol. 1, No. 11, pp. 509-515, 1969.
[13] J. M. SAWHILL, JR., A. W. DIX, W. F. SAVAGE: ‘Modified Implant Test for Studying Delayed
Cracking’, Welding Journal, Vol. 53, No. 12, pp. 554-s-559-s, 1974.

19

View publication stats